Thu 05 Dec 2013 09:28:45 PM UTC, original submission:
Hello,
a forwarded bug from a Debian user: http://bugs.debian.org/731110
--8<--
I observed that wget downloads "wrong" content from a squid cache. Both tools operate within the bounds of RFC2616, but the resulting combination is wrong. Consider the following interaction:
A http client that supports Accept-Encoding and Content-Encoding
downloads an object via a squid3 proxy. An encoding that is not the identity encoding is chosen and the response is considered cacheable by squid3. At a later time wget is instructed to retrieve the same object via the same proxy. Unlike the other client wget does not send an Accept-Encoding header. The squid3 proxy exercises an assumption explicitly granted by RFC2616 14.3:
| If no Accept-Encoding field is present in a request, the server MAY
| assume that the client will accept any content coding.
It thus sends back the cached object including the Content-Encoding header, but wget does not handle the header and stores the object in compressed form.
Note that RFC2616 goes on to say:
| In this case, if "identity" is one of the available content-codings,
| then the server SHOULD use the "identity" content-coding, unless it has
| additional information that a different content-coding is meaningful to
| the client.
So arguably squid3 is violating a SHOULD clause and should be fixed.
A simple mechanism to resolve the situation is to make wget send an empty Accept-Encoding header by default or setting its value to "identity".
Helmut
--8<--
Thank you.
|